Global warming and climate change in recent years have significantly affected the world, and with these has come a significant increase in the frequency and intensity of natural hazards such as landslides, mudflows, and flash floods. These disasters often cause damage to existing infrastructure, which causes a domino effect that hurts the economy and, in worse cases, unavoidable casualty. Having an effective protective structure is of utmost importance to mitigate this. Hence, the behaviour of the protective structure needs to be investigated. The Material Point Method (MPM) is well suited for landslide simulations because of its ability to handle large deformations due to its hybrid approach of an Eulerian framework for the computation grid in combination with Lagrangian moving integration points. MPM can be developed either using fully 3D formulations or depth-averaged approaches. This latter approach allows the simulation of very large domains, which is unfeasible using 3D models. Compared to the Depth-Averaged Method, the strength of a 3D approach is its ability to simulate accurately local effects, such as the impact on a structure, which intrinsically is a very complex phenomenon that can't be simulated in 2D. In this work, we are proposing a multiscale approach by combining the Depth-Averaged Method and a 3D formulation, both using MPM to maximise their strength. The Depth-Averaged Method is used to predict the run-out of a landslide. After determining the location of interest, MPM will be used in a more focused domain to simulate the coupling between an MPM and FEM to model the impact of the landslide on a flexible protective structure. All the codes are implemented in Kratos Multiphysics.